Inductive Voltage Spike

By Tim Daycounter

Inductive loads have a common characteristic that if current is flowing through them and the circuit is interrupted, the inductor will produce a high frequency voltage spike. Current through an inductor doesn't like to be interrupted abruptly, and when a circuit is broken, the result is a very high voltage pulse.

This effect is analogous to a charged capacitor with leads that become shorted. The result is a current spike.

Normally this voltage spike is an unwanted feature, as it tends to destroy MOSFETs or other solid state switches, and switch and relay contacts.

The most common solution is to place a diode (rectifier) in parallel with the inductive load. These diodes are referred to as catch diodes, flyback diodes, clamping diodes, snubber diode, freewheeling diode, or suppressor diodes. Different names for the same action. They are placed in the circuit so that they are reverse biased when the load is energized. When current is cut to the load they provide a return path for the current, so that the current flow through the inductor is not blocked. The current will dissipate over time by their internal resistance, and heat will be generated by the inductor and the rectifier.

If diodes required no time before they started conducting from a reverse biased state, this would be the end of our story, however they require a small bit of time to start conducting. This time is called the forward recovery time. The forward recovery time is not normally listed on a data sheet, but the reverse recover time is listed for nearly all diodes, and they are typically related. A diode is referred to as a fast, or as an ultra fast diode if the reverse recover time is small, on the order of 50ns, and referred to as standard diodes, if this time is slow.

To minimize the inductive voltage spike use fast diodes. The faster the forward recovery time, the smaller the voltage spike, and the shorter its duration. Typically, the fastest diodes you can get are about 20ns. The problem is that even though this 20ns window is short, a large spike can result briefly, if the current is high. This current can be 3 or more times larger than the driving current, and can blow out the driving MOSFET, as they are very sensitive to voltages greater than Vds(max).

To prevent MOSFET failure there are a couple options. You can put some sort of transient surge protector such as a TVS or MOV across the mosfet to absorb these spikes, this requires another component with the associated costs, and they can become hot, especially in situations where the MOSFET is rapidly switching such as with PWM drivers.

Another alternative is to use a higher voltage MOSFET. The problem with this is that there is a trade off between current and voltage on MOSFETs. Typically, the higher the voltage the higher the Rds(on) value, and thus smaller the current for a given packages. The higher voltage FETs tend to be more expensive. So you'll end up spending $3.00 for a FET instead of $0.30 for a lower voltage FET if the inductive voltage spike wasn't an issue.

Snubber networks which consist of passive circuitry which seeks to momentarily store the energy from the inductive voltage spike, or dissipate it, can also be used, but these also are more complex and require more parts and cost, and board real estate.

As a final option, the turn on time of the transistor can be increased with a gate resistor.

Earlier in my career, I would often see MOSFETs with a resistor in series with the gate, and I'd scratch my head and wonder why the resistor had been designed in, thinking that the part was extraneous. My mistake was in thinking that the MOSFET was an ideal component, and that it had an infinite resistance on the gate, and thus the voltage across the gate would be the same as the voltage applied to input side of the resistor, making it unnecessary. Over time I learned that far from being an ideal impedance, the gate of a MOSFET looks like a capacitor. This capacitor tends to be inversely proportional to Rds(on); the higher the current, the slower the MOSFET, and the need for a stronger drive on the gate.

MOSFETS are such great devices because they look almost like ideal switches. They rapidly switch current through them, and normally look like an open or a short with nearly no resistance. As such they are very efficient and dissipate nearly zero power. Because of this they can be used to make very efficient power supplies, or PWM motor controllers. In these sorts of applications, you want the transistor to switch quickly, with as little time spent in the transistors resistive region as possible. To minimize the power dissipated in the FET you want to minimize the gate capacitance, and series resistance to the gate. The trade off is that the more abruptly the MOSFET is switched on, the larger the resulting inductive voltage spike.

Not all applications require rapid switching of MOSFETS, and under these situations, a series gate resistor, can be used to slow down the transition of the the FET. Since the gate looks like a capacitor, a larger series resistor results in a slower charge time of this gate capacitor, and so Vgs, the gate voltage, rises more slowly, and the transistor spends more time in its resistive region, and as such the FET dissipates the current as heat, and little or no voltage spike is formed. Imagine the gate resistor as air bags in your car when you get in a crash. The resistance in the FET as it turns off provides a "cushion" which stops abrupt transitions in current levels.

The trade off with this approach is that while it reduces the inductive current spike, it does increase the power dissipated in the FET, and slows it down. But for many applications, such as those where a FET replaces a mechanical relay, it may not matter.